In June, Bellwether Research and Consulting released a poll examining Hoosiers' views of Gov. Mike Pence and some of his possible opponents in 2016. The immediate reaction was that the reelection bid by Indiana Governor Mike Pence was in trouble.
There were reasons to think this:
- 43% of the respondents had an unfavorable opinion of Pence
- 46% of the respondents disapproved of the job Pence is doing as governor
- 54% of the respondents said they thought it was time to give a new person a chance to do a better job.
If that was not bad enough, in the horse-race question, Pence was losing to John Gregg by one point and was tied with Glenda Ritz.
Not all of the news was bad for Pence:
- The favorable rating for Pence did not change from a Bellwether poll in April.
- The unfavorable ratings for Ritz and Gregg went up from April.
- The apparent changes in the horse-race questions were within the margin of error and so may not have changed at all.
- A plurality of respondents wanted a Republican to be elected governor.
- 53% of the respondents approved of how Pence was handling the economy.
It is important when a poll like this comes out for us to remember a couple of things. First, a poll is a snapshot in time. The results found by the poll are not inevitable (e.g. these results do not mean that Gregg will win by one point.) Second, the poll is a piece of information that needs to be considered in conjunction with other information.
Polls like this one provide an opportunity to remember how to be informed consumers of polling data. In simple terms good polls ask the appropriate people the questions, ask good questions, and get analysis from a reliable source.
Conducting a poll does not require speaking with a large number of people. Only 800 people responded to this poll and that resulted in a margin of error of plus or minus 3.5 points. That means the results you read could vary by 3.5 points in either direction. In other words, the unfavorable percentage for Pence could be as low as 39.5% or as high as 46.5%.
A good sample starts with identifying the population. In the case of a poll about an election, the population is likely voters. It might sound undemocratic, but in the end, the voices that matter in an election are the voices of people who are going to show up and vote. Once the population is known, then we have to find a way to make sure it is representative of the population. In election polling, variables like gender, age, educational attainment, and party identification may be used. Another variable that might be used in a statewide poll is geography (rural, suburban, and urban).
Writing good questions is not as easy as it sounds.
Can the respondents reasonably be expected to answer the question? We can expect people to be able to tell us their favorite dessert, but can we expect people to tell us what will happen if there is a reduction in an inventory tax?
Is the question clear to the respondent? The use of double negatives or unconventional wording can confuse the respondents and throw off the results.
Even if the respondent can be expected to answer the question and the wording is clear, the wording of the question still matters. In the summer of 2008, the Mike Downs Center did four polls in Indiana asking voters who they would support for governor later that year. Each survey asked the question in a different way and the results ranged from a tie to a race that was safely in Mitch Daniel’s hands.
The final element of a good poll is good analysis.
It is best if the analysis is being done by someone who analyzes political polls. Understanding voter behavior is different than understanding shopping habits. Even if the analyst focuses on political polling, some analysts are better than others. In this case, Christine Matthews is quite good and her analysis deserves attention.
When it comes to analysis, we should ask if the conclusions drawn by the analyst logically can be drawn from the survey. For example, in this case it is difficult to draw any conclusions about a contested Republican primary, because the only question about the Republican primary is whether or not the respondent will vote in the primary election.
All of this leads to the question of what this poll tells us about the 2016 Indiana gubernatorial election. The answer is that the results are interesting, provide some insight into what voters in Indiana are thinking right now, and what candidates might want to do. But the results tell us little about the likely outcome of the 2016 gubernatorial election.
It is June of 2015. Election Day 2016 is November 8, 2016. Far too many things can happen between now and then for us to think these results are written in stone.
Andrew Downs is Director of the Mike Downs Center for Indiana Politics at IPFW.
Opinions expressed in this column are those of the individual writer and do not necessarily reflect the opinions of the staff, management or board of Northeast Indiana Public Radio. If you want to join the conversation, head over to our Facebook page and comment on the post featuring this column.